Inflexible Interpretations of Ambiguous Social Situations: A Novel Predictor of Suicidal Ideation and the Beliefs That Inspire It

Abstract

Suicidal ideation has been linked to a bias toward interpreting ambiguous information in consistently less positive or more negative manners (positive/negative interpretation bias), implying that information-processing biases might distort beliefs thought to inspire suicidal ideation (e.g., those regarding burdensomeness). Therefore, in the present study, we examined whether suicidal ideation and beliefs highlighted in theories of suicide are related to positive/negative interpretation bias and/or a bias against revising negative interpretations in response to evidence against them (negative interpretation inflexibility). Data were collected in three waves, each 1 week apart. Network analyses and structural equation models provided evidence that negative interpretation bias (cross-sectionally) and negative interpretation inflexibility (cross-sectionally and over time) were related to suicidal ideation and that the latter relationship was mediated by perceived burdensomeness. By identifying this mediation pathway in the present study, we provide a potential mechanism by which perceptions of burdensomeness, a key risk factor for suicidality, might arise and/or persist.

Keywords

suicidal ideation interpretation bias interpretation inflexibility perceived burdensomeness open data preregistered

Suicide causes approximately 800,000 deaths annually (World Health Organization, 2019), and predicting who will attempt or die by suicide has proven challenging (Franklin et al., 2017). These facts imply that the current understanding of suicidality is inadequate for the development of sufficiently effective suicide-prevention strategies. Although most studies meant to help remedy this situation have investigated suicide-related behaviors (e.g., suicide attempts), a growing number have examined suicidal ideation (as called for by suicide researchers, see Jobes & Joiner, 2019), which involves plans or wishes to die by suicide without overt action (Beck et al., 1979) and is an important step along pathways leading to suicide (Jobes & Joiner, 2019). These studies demonstrate that suicidal ideation is strikingly common (Jobes & Joiner, 2019), can cause psychological distress, and sometimes interferes with daily life (Van Spijker et al., 2014).

These studies also establish an association between suicidal ideation and inflexibility in multiple psychological processes (e.g., coping flexibility: Heffer & Willoughby, 2017; difficulties with set-shifting: Jyunn Lai et al., 2018; emotion regulation: Rogers & Joiner, 2017). Inflexibility in these processes may play important roles in pathways leading to suicidal ideation. Consistent with this notion, inflexibility in set shifting prospectively predicts suicidal ideation through a plausible mediating mechanism (increasing brooding and, in turn, hopelessness; Miranda et al., 2012, 2013) and exhibits a dose-response relation with suicidal ideation (Jyunn Lai et al., 2018). Given the potential importance of inflexibility in pathways leading to suicidal ideation, investigations of whether inflexibility in other psychological processes is linked to suicidal ideation and, if so, by what potential mechanism or mechanisms are warranted. In the present study, we therefore investigated the relation between suicidal ideation (as well as its known risk factors) and interpretation inflexibility (failure to adequately update interpretations in response to changing situational demands), a specific form of belief inflexibility (the metacognitive ability to reflect on one’s beliefs and consider alternatives in light of the possibility they may be mistaken; Ward et al., 2018).

Although belief inflexibility has been underexamined in studies of suicidal thoughts (and behaviors), the potential relevance of belief inflexibility to suicidal ideation is evident when considering that theories of suicidality have consistently hypothesized that suicidal ideation is influenced by individuals’ beliefs about their present and future circumstances. The interpersonal theory of suicide (Joiner, 2005), for instance, asserts that beliefs regarding thwarted belongingness (a mental state resulting from unmet needs for connectedness) or perceived burdensomeness (a mental state characterized by [mistaken] beliefs that “others would be better off if I were gone”) result in passive suicidal ideation (e.g., “I wish I were dead”), which escalates to active suicidal desire (e.g., “I would kill myself if I had the chance”) if both types of beliefs are present and the interpersonal situation that begot these beliefs is deemed intractable (Chu et al., 2017). Moreover, the three-step theory of suicide (Klonsky & May, 2015) posits that suicidal ideation results when individuals experience pain (most often, psychological or emotional pain, which is referred to as psychache) and endorse beliefs about their future characterized by hopelessness (expecting negative future events one is helpless to influence; Abramson et al., 1989). Finally, the integrated motivational-volitional model of suicidal behavior (R. C. O’Connor, 2011) attributes a critical role to perceptions of defeat (failed social struggle) and entrapment (perceived inability to escape aversive situations) in the development of suicidal ideation. Specifically, the model predicts that suicidal ideation is caused by defeating or humiliating experiences from which one believes there is no escape.

The central role beliefs play in these prominent accounts of suicidal ideation is notable because beliefs often systematically deviate from reality because of biases in information processing (e.g., confirmation bias; Nickerson, 1998). This is particularly true in the context of psychopathologies (e.g., depression; see Everaert, Bernstein, et al., 2020; Everaert et al., 2017) that frequently feature suicidal ideation (Vuorilehto et al., 2014). Unsurprisingly, then, research indicates that beliefs that might inspire suicidal ideation are not always congruent with reality. For example, in the context of suicidal thoughts and behaviors, perceptions of burdensomeness are most often thought to be inaccurate (Hames et al., 2013), a position consistent with interview-based studies indicating that these perceptions are often fueled by normative experiences (e.g., in the case of adolescents: needing transportation by close others; Buitron et al., 2016).

Beliefs regarding burdensomeness (along with other beliefs that putatively inspire suicidal ideation) may become incongruent with reality as a result of biased interpretation (the semantic process of integrating aspects of a situation to resolve ambiguity; Blanchette & Richards, 2010). Consistent with this possibility, preliminary evidence suggests that cognitive bias modification programs designed to ameliorate interpretation bias can reduce perceived burdensomeness and suicidal ideation (Allan et al., 2018). As noted by Beard et al. (2017), interpretation bias may also influence beliefs about individuals’ futures in a manner that promotes hopelessness and suicidal ideation. This possibility is strongly consistent with the fact that negative interpretation bias prospectively predicts hopelessness and suicidal ideation (Beevers & Miller, 2004). Interpretation bias may distort these beliefs by interfering with the ability to accurately decipher social environments, which often feature ambiguous situations requiring interpretation (Everaert et al., 2018).

Above and beyond any effect or effects of interpretation bias, interpretation inflexibility might promote additional discrepancies between reality and beliefs that inspire suicidal ideation. Consistent with this notion, inflexible negative interpretations may result in maladaptive beliefs about positive emotions (e.g., “good things won’t last”), leaving individuals vulnerable to experiencing depressive symptoms (Everaert, Bronstein, et al., 2020), which often include suicidal ideation (Vuorilehto et al., 2014). Inflexible negative interpretations may also maintain overly negative appraisals of ambiguous interactions with others (potentially increasing perceived burdensomeness and thwarted belongingness) and cause individuals to discount evidence that their problems have improved (increasing hopelessness), although these possible effects of inflexible interpretations have not been evaluated in extant literature.

The Present Study

The literature reviewed above suggests that inflexible/biased interpretations may cause beliefs to deviate from reality in a manner that prompts suicidal ideation. In the present study, we therefore examined whether inflexible and biased interpretations are related to suicidal ideation via their impact on beliefs about individuals’ current and future circumstances, such as those related to perceived burdensomeness and hopelessness. This examination was conducted in two stages. First, in the tradition of several previous studies of suicidal thoughts and behaviors (Bloch-Elkouby et al., 2020; de Beurs et al., 2019), cross-sectional network analyses were employed with the goal of uncovering potential pathways by which inflexible and biased interpretations may be related to suicidal ideation and the beliefs that putatively inspire it. Second, structural equation models were used to examine whether potential mediation pathways identified in this analysis held in a longitudinal data set. Modeling longitudinal data was necessary to establish whether inflexible interpretations are a risk factor for suicidal ideation; by definition, risk factors must precede and predict their associated outcomes. The test of mediation within these models was included to facilitate identification of plausible mechanisms temporally linking inflexible interpretations and suicidal ideation. Identifying these mechanisms is an important step toward determining whether inflexible interpretations may be a manipulative cause of suicidal ideation (see Cox, 2018; Lervåg, 2019).

Two hypotheses were tested via these analyses:

Hypothesis 1: Negative interpretation inflexibility and interpretation bias will be independently associated with suicidal ideation and with beliefs highlighted in major theories of suicide (e.g., the belief that one is a burden to others), and these beliefs will mediate any relation between suicidal ideation and interpretation inflexibility or bias.

Hypothesis 2: Negative interpretation inflexibility and interpretation bias will independently predict longitudinal increases in suicidal ideation and the endorsement of beliefs related to suicide (e.g., hopelessness).

Note that these hypotheses were investigated using the emotional bias against disconfirmatory evidence (BADE) task (Everaert et al., 2018). This task’s key strength, the ability to disentangle the effects of interpretation bias and inflexibility, allowed us to examine the unique relations between inflexible and/or biased interpretations and suicidal ideation. This capability is beneficial because people are generally more reluctant to revise beliefs and interpretations in which they are more confident (Kaliuzhna et al., 2012; Woodward et al., 2008). This tendency implies that failure to account for interpretation bias, which might inspire greater confidence in interpretations, could potentially confound any test of the relation between interpretation inflexibility and suicidal ideation. Accordingly, the emotional BADE task’s ability to separate interpretation bias and inflexibility allows us to rigorously test the foregoing hypotheses.

Method

Participants and recruitment

Participants (Phase 1 n = 853, completed Phase 2 n = 207; for demographics, see Section S2 in the Supplemental Material available online) were recruited in two phases using a stratified random sampling approach to ensure adequate variability in suicidal ideation across individuals. Phase 1 served as a prescreening. Phase 2 was the main study (for attrition data, see Section S4 in the Supplemental Material). Participants were Amazon Mechanical Turk (MTurk) workers. MTurk provides an online crowdsourcing platform with access to large and diverse samples suitable for clinical research collecting mental health data (Chandler & Shapiro, 2016), including studies investigating individual differences in suicidal ideation (Shahnaz et al., 2018). Participation was limited to MTurk workers age 18 years and older living in the United States. Participants received $9 (USD) remuneration.

Data-quality measures

Following recommendations for research using crowdsourced samples (Chandler & Shapiro, 2016), several steps were taken to ensure high data quality. Workers were required to have a history of providing good-quality responses (i.e., an acceptance ratio of ≥ 96% in a minimum of 100 human intelligence tasks), to complete a captcha (to deter “participation” by computer programs), to pass five of six included attention-check questions (placed at irregular intervals to identify inattention), and to have an IP address and MTurk ID that were not associated with low-quality data or a non-U.S. location (Ophir et al., 2018; Prims et al., 2018). Participants self-reporting that they answered Phase 1 survey questions randomly or did not follow survey instructions were not invited to participate in Phase 2. Survey measures were checked for adequate internal consistency using McDonald’s (1999) ω total; all measures had good or excellent internal consistency across all data collection waves (see Section S4 in the Supplemental Material). These data-quality measures adhere to current standards and are strongly based on those that have yielded highly replicable results in past research (e.g., Bronstein et al., 2019; Everaert, Bronstein, et al., 2020; Everaert et al., 2018). Nevertheless, replication of the present study is encouraged because this is the best way to ensure the reliability of its conclusions.

Protocol

The Yale Institutional Review Board approved the study. In Phase 1, 853 MTurk workers completed the five-item Scale for Suicide Ideation (SSI-5; the first five items of the Beck Scale for Suicide Ideaton: Beck & Steer, 1991), which determined invitations to Phase 2. As the first step in the invitation process, SSI-5 scores were recoded into three groups. Group 1 reported no suicidal ideation (SSI-5 = 0). The remaining groups were divided via a median split of all SSI-5 scores exceeding zero (Group 2: M = 1.52, SD = 0.50; Group 3: M = 4.61, SD = 1.77). From each group, 125 individuals selected at random received invitations for Phase 2 (this choice was informed by a power analysis suggesting that 375 participants would be sufficient to detect even a small effect of interpretation inflexibility on suicidal ideation; see https://osf.io/hf4e8).

Phase 2 included a baseline assessment plus two follow-up assessments (after 1 and 2 weeks, respectively). At baseline only, participants completed the emotional BADE task (Everaert et al., 2018). During all three sessions, participants completed the Beck Depression Inventory–II (Beck et al., 1996), SSI-5 (Beck et al., 1979), Beck Hopelessness Scale (Beck et al., 1974), Interpersonal Needs Questionnaire (Van Orden et al., 2012), Scale of Psychache (Holden et al., 2001), Perceived Stress Scale (Cohen, 1994), Short Defeat and Entrapment Scale (SDES; Griffiths et al., 2015), Future Events Questionnaire (FEQ; Miranda & Mennin, 2007), and the Future-Oriented Repetitive Thought–Pessimistic Repetitive Future Thinking Scale (Miranda et al., 2017). Within each session, measures were presented in randomized order. Participants completed questionnaires in reference to the past week to standardize temporal orientation across questionnaires and data collection waves.

Measures: suicidal desire and psychiatric history

Current suicidal desire

Suicidal desire was measured using the SSI-5. The SSI-5 has five items (measuring wish to live, wish to die, reasons for living/dying, desire to make an active suicide attempt, and passive suicidal desire) that capture the construct of “suicidal desire.” For each item, participants reported which of three potential responses, organized by increasing severity, was most applicable to them. The SSI-5 was derived from the SSI (Beck et al., 1979), a 19-item instrument evaluating the presence and intensity of suicidal thoughts over the past week. Previous research supports the five-item version of the questionnaire (de Beurs et al., 2015). The SSI-5 had good to excellent internal consistency in the present study (minimum: ω_total = .86).

Psychiatric history

Lifetime history of psychiatric diagnosis, past and current help-seeking behavior, lifetime history of suicide attempt, and the lethality of most severe suicide attempt (if any), were queried (see Section S3 in the Supplemental Material).

Measures: established risk factors for suicidal ideation

Depression symptoms

The BDI-II (Beck et al., 1996; present study: minimum ω_total = .96) includes 21 items measuring depressive symptom severity. Each item queries the degree to which individuals have experienced a certain symptom on a 4-point scale (range = 0–3). Total scores are calculated as the sum of these ratings. However, Item 9 (which examines suicidal desire) was excluded except when reporting descriptive statistics and zero-order correlations (to avoid redundancy with the SSI-5; analyses used herein assume lack of redundancy; see Meier et al., 2019). The BDI-II has good reliability and validity (Dozois et al., 1998). Depression is a risk factor for suicidal ideation and behavior (Henriksson et al., 1993; Isometsä, 2014) and may mediate the relation between other risk factors and suicidal ideation (Wang et al., 2015).

Hopelessness

The Beck Hopelessness Scale (Beck et al., 1974; present study: minimum ω_total = .93) contains 20 items measuring individuals’ negative attitudes and expectations about the future and amotivation. Individuals indicate whether self-report items are true or false for them. This measure has good internal consistency and validity (Kocalevent et al., 2017). Hopelessness is central to important theories of suicide (e.g., Abramson et al., 2006) and has been repeatedly related to suicidal ideation (e.g., Dhingra et al., 2019; Klonsky & May, 2015).

Burdensomeness and belongingness

The Interpersonal Needs Questionnaire (Van Orden et al., 2012; present study: minimum ω_total = .96) has subscales assessing thwarted belongingness (i.e., beliefs about the extent to which individuals feel connected to others) and perceived burdensomeness (i.e., degree to which individuals feel they are a burden to others). Respondents rate the INQ’s 15 items using a 7-point scale. The questionnaire has strong psychometric properties (Van Orden et al., 2012). Perceived burdensomeness and thwarted belongingness are the central constructs in the interpersonal theory of suicide (Joiner, 2005); extensive evidence links them to suicidal ideation (Chu et al., 2017).

Psychological pain

The 13-item Scale of Psychache (Holden et al., 2001; present study: minimum ω_total = .97) measures current emotional or mental pain. Respondents rate whether they experience various aspects of psychache on a 5-point Likert scale (1 = never, 5 = always). This scale has favorable psychometric properties (Holden et al., 2001). Psychache features prominently in several theories of suicide (Klonsky & May, 2015; Shneidman, 1993) and has been consistently related to suicidal ideation (Verrocchio et al., 2016).

Perceived stress

The Perceived Stress Scale (Cohen et al., 1983; present study: minimum ω_total = .92) includes 10 items measuring whether situations are appraised as stressful (i.e., unpredictable, uncontrollable, and overwhelming). Items are rated on a 5-point scale (0 = never, 4 = very often). The questionnaire has adequate internal consistency (Roberti et al., 2006). In the integrated motivational-volitional model of suicidal behavior, stress is thought to combine with preexisting vulnerabilities (e.g., in biological stress-response systems, see: Miller & Prinstein, 2019) to increase the likelihood that individuals will consider suicide (R. C. O’Connor & Kirtley, 2018). Consistent with this notion, multiple studies have observed associations between suicidal ideation and acute stress (e.g., Cole et al., 2015; D. B. O’Connor et al., 2017).

Defeat and entrapment

The Short Defeat and Entrapment Scale (SDES; Griffiths et al., 2015; present study: minimum ω_total = .94) is an eight-item measure of defeat and entrapment. Each item is rated on a scale from 0 (not at all like me) to 4 (extremely like me). Defeat and entrapment were treated as one construct given previous research suggesting that they load onto one factor (Sturman, 2011; Taylor et al., 2009; but see Forkmann et al., 2018). Defeat and entrapment are central to the integrated motivational and volitional model of suicidal behavior (R. C. O’Connor et al., 2016; R. C. O’Connor & Kirtley, 2018); extant literature strongly links these constructs to suicidal ideation (R. C. O’Connor & Portzky, 2018).

Confident anticipation of negative and (few) positive future events

The Future Expectancies Questionnaire (Miranda & Mennin, 2007; present study: minimum ω_total = .90) was used to measure confidence in expectations regarding future negative and positive events. Items ask whether 34 future events (half positive/negative) will happen (yes/no), along with confidence in that prediction (on a 5-point scale; 1 = not at all certain, 5 = as certain as one can be). This questionnaire has adequate internal consistency (Krajniak et al., 2013; Miranda & Mennin, 2007). As measured by this questionnaire, the confident lack of positive future expectancies is associated with suicidal ideation even when controlling for hopelessness and depression (Sargalska et al., 2011). This questionnaire was included because the integrated motivational-volitional model of suicidal behavior posits that beliefs about the future may influence the relation between entrapment and suicidal ideation (R. C. O’Connor & Kirtley, 2018).

Pessimistic repetitive future thinking

The Pessimistic Repetitive Future Thinking subscale of the Future-oriented Repetitive Thought Scale The Pessimistic Repetitive Future Thinking subscale of the Future-oriented Repetitive Thought Scale (Miranda et al., 2017; present study: minimum ω_total = .94) measures repeated thought about whether negative future outcomes will occur or positive future outcomes will not occur. Respondents rate eight items (0 = never, 4 = almost always). This subscale has proven reliable in several studies (Miranda et al., 2017). Pessimistic repetitive future thinking has been associated with suicidal ideation and is elevated in individuals with a history of suicide attempts (Kirtley et al., 2018). This construct lies at the intersection of rumination (which the integrated motivational-volitional model posits influences the relation between defeat and entrapment) and future thinking (which this model posits modulates the relation between entrapment and suicidal ideation). Pessimistic repetitive future thinking was measured because it may alter the character of individuals’ beliefs about their futures (e.g., by increasing their certainty about predicted negative events via rehearsal), thereby encouraging suicidal ideation (Krajniak et al., 2013).

Measures: interpretation bias and inflexibility

Emotional bias against disconfirmatory evidence task

Interpretation bias and inflexibility were examined using the BADE task (Everaert et al., 2018). During the emotional BADE task, self-referential scenarios containing three statements were presented to participants (for an example, see Section S9 in the Supplemental Material). Each statement provided additional information about an unfolding ambiguous interpersonal situation containing themes of social failure and rejection. After viewing each statement, participants rated the plausibility of four different interpretations of the information acquired thus far in that scenario using a 21-point rating scale ranging from 1 (poor) to 21 (excellent). Interpretations were presented in randomized order after each statement.

Across scenarios, interpretations were grouped into three categories: absurd (e.g., “You and your old friends suddenly fall asleep”), lure (e.g., “Your old friends decide to leave and do something else”), and true (e.g., “Your old friends think you are a fun person to spend time with”) interpretations (in accordance with Everaert et al., 2018). Absurd interpretations were consistently implausible throughout each scenario. Lure interpretations (Lure-A and Lure-B; lure interpretations are arbitrarily sorted into these categories) in each scenario were initially the most plausible but became less so as the scenario continued. True interpretations were initially only moderately plausible but became the most plausible interpretation by the end of the scenario. Given this structure, optimal performance on the emotional BADE task requires participants to revise beliefs about the most plausible interpretation for a given scenario by integrating the disconfirmatory information provided by each of the latter two statements.

The emotional BADE task includes two types of scenarios: disconfirming-the-negative scenarios (which contain negative lures and a positive true interpretation, inviting revision of negative interpretations in response to disconfirmatory positive information) and disconfirming-the-positive scenarios (which contain positive lures and a negative true interpretation, inviting revision of positive interpretations in response to disconfirmatory negative information). Given previous research (e.g., Everaert et al., 2018) suggesting that metrics of interpretation inflexibility derived from disconfirming-the-negative scenarios may be more strongly related to depressive symptoms, potentially including suicidal ideation, results regarding data from these scenarios were of central interest herein (for results from disconfirming-the-positive scenarios, see Section S8 in the Supplemental Material).

Three metrics were derived from disconfirming-the-negative scenarios: negative interpretation inflexibility, negative interpretation bias, and positive interpretation bias. In previous studies (Everaert, Bronstein, et al., 2020; Everaert et al., 2018), these metrics were calculated using principal component analysis (PCA). Using PCA to score the emotional BADE task was advantageous in revealing important directions of variance in task performance. However, this method has noteworthy disadvantages (for a review, see Bronstein & Cannon, 2018). Researchers have therefore recommended that the original BADE task (upon which the emotional BADE task was based) be scored using alternative methods (Bronstein & Cannon, 2018; Sanford et al., 2014). One recommended strategy is to calculate BADE task metrics as unweighted sums of the average explanation ratings loading most strongly onto each PCA-derived BADE component (Sanford et al., 2014). In the present study, we applied this strategy to the emotional BADE task for the first time. As noted in the preregistration for the present study, each metric was scored as follows: negative interpretation inflexibility (Absurd1 + Absurd2 + Absurd3 + LureA3 + LureB3 − True3), positive interpretation bias (True1 + True2 + True3), and negative interpretation bias (LureA1 + LureB1 + LureA2 + LureB2). In these formulas, numbers in variable names represent the statement after the rating was made, and the remainder of the name denotes the category of explanation being rated.

Analyses

The primary analyses in the present study (those involving network analysis and longitudinal structural equation modeling) were preregistered (https://osf.io/hf4e8).

Outliers and missing data

Outliers were detected using the method of Hubert and Van Der Veeken (2008), as preregistered, because this method was designed for detecting outliers in skewed data; several variables in the present study (interpretation inflexibility, suicidal ideation) were right skewed. Detected outliers were winsorized (Fuller, 1991). Analyses of cross-sectional data were conducted after removing participants with missing data listwise. Analyses of longitudinal data were conducted with these participants included using proper missing data handling techniques (see below).

Hypothesis 1: cross-sectional associations between suicidal ideation, suicide-related beliefs, and interpretation bias/inflexibility

In community samples, individuals often deny suicidal ideation, giving it a right skewed distribution. For example, 56% of all individuals (137 of 244) denied suicidal ideation during the baseline session of Phase 2. The large number of zeros in this distribution may cause commonly employed regression models (e.g., multiple linear regression or Poisson regression) to produce unstable coefficients for predictors, potentially resulting in Type I errors. To overcome this disadvantage, research on suicidal ideation (e.g., Chu et al., 2018; Cukrowicz et al., 2013) has begun to employ zero-inflated negative binomial regression models (Greene, 2012; for additional information, see Section S6 in the Supplemental Material).

In the present study, zero-inflated negative binomial regression models—implemented via the pcsl package (Version 1.5.5; Jackman, 2020) for the R software environment (Version 3.6.1; R Core Team, 2019)—were used to test the portion of Hypothesis 1 asserting that negative interpretation inflexibility and negative/positive interpretation bias would be uniquely associated with suicidal ideation. Predictors were mean-centered and z scored. The criterion was left in raw units to maintain its status as a count variable. Bias-corrected accelerated 95% confidence intervals (CIs) for model coefficients were generated with 1,000 resamples. Data from the Phase 2 baseline were modeled. Depression was statistically controlled because it associates with interpretation inflexibility and suicidal ideation.

Hypothesis 1 was then tested more fully in these data using network analysis to simultaneously analyze and visualize complex relations between suicidality and its many risk factors (de Beurs, 2017). The network analysis in the present study was conducted in three stages (readers less familiar with network analysis, see Section S10 in the Supplemental Material).

Network estimation

First, network structures were estimated using measures of suicidal ideation, depression, perceived burdensomeness, thwarted belongingness, psychache, defeat/entrapment, hopelessness, certain positive and negative future expectancies, pessimistic repetitive future thinking, perceived stress, positive/negative interpretation bias, and interpretation inflexibility. Partial correlations were used to calculate edges. The least absolute shrinkage and selection operator (LASSO; Tibshirani, 2016) was used to regularize the resulting network. The tuning parameter for LASSO (λ) was selected to minimize the extended Bayesian information criterion (EBIC; Chen & Chen, 2008) because this technique is beneficial for retrieving the true network structure (Foygel Barber & Drton, 2015). Following the recommendations of Foygel and Drton (2010), the EBIC hyper-parameter (γ) was set to .5 to prioritize avoidance of Type I errors. The resulting undirected graphs of regularized partial-correlation networks were visualized using the qgraph package (Version 1.6.5; Epskamp et al., 2020) for the R software environment.

Because network analysis assumes that nodes represent distinct entities (Meier et al., 2019), variables included in each network analyses were evaluated for potential redundancy using the networktools package (Version 1.2.2; Jones, 2020) for R. In keeping with prior research, a given pair of variables was considered potentially redundant if (a) zero-order correlation between the variables exceeded .70 (indicating at least 50% overlapping variance; following the threshold adopted in Elliott et al., 2020) and (b) correlations between each member of the pair and all remaining variables in the network were not statistically significantly different in more than 75% of cases (Marchetti, 2020; Meier et al., 2019).

Network inference

After the network was estimated, node centrality and predictability were examined. Node predictability was computed using the mgm package (Version 1.2.7; Haslbeck, 2019) for R. Predictability indexes how much variance in a given node is explained by the nodes that are connected to it in the network (Haslbeck & Fried, 2017) and can be understood as an upper bound on controllability (Fried et al., 2018).

Node centrality was measured using one-step expected influence (in networks with positive and negative edges, this measure of centrality may be preferable; see Robinaugh et al., 2016). One-step expected influence was calculated as the sum of the value of all edges extending from a given node, taking the mathematical sign of each edge into account. Expected influence was calculated using networktools (Jones, 2020).

Because it is possible that unequal variances of the nodes in the network affect their centrality estimate, thereby influencing the observed network structure (Terluin et al., 2016), correlations between expected influence values and standard deviations of the individual nodes were examined. The correlations between expected influences and means of individual nodes were also examined to better ensure that differences in severity did not explain node centrality or network structure.

Statistical tests of the differences between edge weights and node centrality measures were carried out using a nonparametric bootstrapping procedure conducted using the bootnet package (Version 1.3; Epskamp, 2020) for R. Data were bootstrapped 1,000 times for this purpose. If the 95% CI for the difference between two edge weights or node centrality metrics did not overlap with zero, the difference was considered statistically significant.

Network accuracy and stability

Network accuracy and stability were also estimated using bootnet and 1,000 bootstrapped resamples. The accuracy of edge weights was quantified using the 95% CIs generated from this bootstrapping procedure. Narrower CIs indicate greater precision in edge weight estimates. Stability of centrality measures was quantified using the correlation stability coefficient (CS-coefficient; Epskamp et al., 2018), which denotes the proportion of cases that can be dropped such that the set of stability measures obtained using the full and reduced data sets are correlated above a certain threshold (.70 in the present study) with 95% probability. Following the recommendations of Epskamp et al. (2018), metrics were considered stable if the CS-coefficient exceeded .25.

Hypothesis 2: Do suicide-related beliefs longitudinally mediate the relation between inflexible/biased interpretations and suicidal ideation?

Hypotheses 1 and 2 were then examined simultaneously in longitudinal data from Phase 2. Specifically, these data were examined to determine whether positive/negative interpretation bias and interpretation inflexibility were related to future levels of suicidal ideation and whether these relations were mediated by beliefs highlighted in major theories of suicide. Path models were specified using the lavaan package (Version 0.6-5; Rosseel, 2019) for R. Maximum likelihood with robust standard errors—a variant of the maximum likelihood approach that is robust to nonnormality—was used to estimate parameters and associated standard errors. Each model (see Section S7 in the Supplemental Material) contained negative interpretation inflexibility and positive/negative interpretation bias as exogenous variables (measured at baseline), suicidal ideation as an endogenous variable (measured at Week 2), and constructs from major theories of suicide (e.g., perceived burdensomeness and thwarted belongingness) along with depression as parallel mediators (measured at Week 1). Each path model controlled for baseline levels of the mediator or mediators, depression, and suicidal ideation. Full information maximum likelihood was used to handle missing data (Enders & Bandalos, 2001). Model fits were examined using the following indices (and cutoffs): comparative fit index (CFI; > .95), root mean square residual (RMSEA; < 0.06), and standardized root mean square residual (SRMR; < .08; Hu & Bentler, 1999). χ² statistics were not considered because they tend to overreject models in the presence of large sample sizes (Bentler & Bonett, 1980). Bias-corrected bootstrapped 95% CIs with 1,000 resamples were generated for all parameters; intervals that do not contain zero indicate significant effects. Indirect effects were deemed significant if Monte Carlo simulation—using the semTools package for R, monteCarloMed function with 1,000,000 replications (Version 0.5.2; Jorgensen et al., 2019)—produced CIs that did not contain zero, as recommended by MacKinnon et al. (2002).

Results

Descriptive statistics

Significant variation in suicidal ideation was observed among participants entering study Phase 2. Although the majority of these individuals reported no current ideation (SSI-5 = 0: 137), a substantial number endorsed at least some ideation (SSI-5 = 1: 24; SSI-5 = 2: 24; SSI-5 = 3: 14; SSI-5 = 4: 13; SSI-5 = 5+: 32). The level of suicidal ideation among individuals entering Phase 2 (M = 1.51, SD = 2.27) lay between that observed in general population samples (M = 0.22, SD = 0.98; Kliem et al., 2017) and samples of suicide attempters (M = 2.18, SD = 2.76; Desseilles et al., 2012). Fifty-six Phase 2 participants reported a lifetime history of suicide attempt (186 denied this history). For additional descriptive statistics, see Table S4 in the Supplemental Material.

Zero-order correlations between study variables

Zero-order correlations between variables were calculated at Phase 2 baseline (see Table S5 in the Supplemental Material). Nonparametric (Spearman’s ρ) correlations were employed to account for nonnormality. Note that negative interpretation inflexibility was associated with suicidal ideation, ρ(242) = 0.23, p < .001, and with several suicide-related beliefs (e.g., perceived burdensomeness, thwarted belongingness, and defeat/entrapment). Negative interpretation bias, ρ(242) = 0.09, p = .146, and positive interpretation bias, ρ(242) = −0.07, p = .301, were not associated with suicidal ideation. Despite this, negative interpretation bias was associated with several constructs from prominent theories of suicide, for example, hopelessness, ρ(242) = 0.20, p = .001.

Multiple regression models

To examine the portion of Hypothesis 1 asserting that suicidal ideation would be uniquely associated with interpretation inflexibility and interpretation bias, we constructed zero-inflated negative binomial multiple regression models using data from Phase 2 at baseline. Suicidal ideation was the criterion variable. Interpretation inflexibility, positive/negative interpretation bias, and depression were simultaneous predictors.

As a preliminary analysis, Vuong’s (1989) test was used to examine whether a zero-inflated model was merited. Results indicated that this model fit the data better than a standard negative binomial regression model—implemented using the MASS package for R (Version 7.3-51.6; Ripley, 2020), z = 3.18, p = .001. Accordingly, the results of the zero-inflated and count models (Table 1) were interpreted.

Table 1.

Zero-Inflated Negative Binomial Regression Model With Suicidal Ideation as the Criterion Variable: Results for Disconfirming-the-Negative Scenarios

Predictor	Estimate	SE	z	p	Odds ratio
Negative binomial regression (count model)
NII	0.14 [−0.04, 0.25]	0.08	1.82	.070
NIB	−0.20 [−0.35, −0.01]	0.08	−2.38	.017
PIB	−0.12 [−0.25, 0.07]	0.08	−1.53	.123
BDI^a	0.40 [0.19, 0.52]	0.08	4.83	< .001
Intercept	0.81 [0.70, 1.10]	0.11	7.20	< .001
Logistic regression (zero-inflation model)
NII	−0.48 [−0.94, 0.07]	0.25	−1.93	.054	0.62
NIB	−0.06 [−0.56, 0.53]	0.24	−0.23	.817	0.94
PIB	0.28 [−0.24, 0.82]	0.24	1.17	.243	1.32
BDI^a	−1.53 [−1.91, −0.92]	0.30	−5.08	< .001	0.22
Intercept	−0.12 [−0.37, 0.51]	0.25	−0.48	.629

Note: All variables were measured during the baseline session of study Phase 2. Values in brackets are bias-corrected accelerated 95% confidence intervals. Boldface type indicates significant results. NIB = negative interpretation bias; PIB = positive interpretation bias; NII = negative interpretation inflexibility; BDI = Beck Depression Inventory (Beck et al., 1996).

Item 9 of the BDI, which asks about suicidal thoughts/wishes, was omitted to avoid redundancy with the five-item Scale for Suicide Ideation.

The count model indicated that the baseline level of suicidal ideation was 0.81 among individuals with the potential to report ideation (i.e., among potential ideators). Depression and negative interpretation bias were statistically significant predictors of the amount of ideation reported among these individuals. Model coefficients suggested that there would be a 55% increase (25% decrease) in suicidal ideation if depression increased (decreased) from mean levels to 1 SD above (below) the sample mean and that an analogous increase (decrease) in negative interpretation bias would result in a 30% decrease (45% increase) in suicidal ideation. Negative interpretation inflexibility fell just outside the conventional threshold for significance as a predictor of the amount of suicidal ideation reported by potential ideators. Its associated coefficient suggested that increasing (decreasing) negative interpretation inflexibility by 1 SD from mean levels could potentially result in a 42% increase (32% decrease) in suicidal ideation.

In the zero-inflated model, the intercept was not significantly different from zero (i.e., the baseline odds of not reporting zero suicidal ideation did not significantly differ from 50%). Depression was a statistically significant predictor of whether a case would be an excess zero (nonideator), whereas negative interpretation inflexibility again fell just outside the conventional threshold for significance. The coefficients suggested that a 1 SD increase (decrease) in depression above (below) its mean would portend a 78% decrease (362% increase) in the conditional odds of being a nonideator. Furthermore, the coefficients suggested that a 1 SD increase (decrease) in negative interpretation inflexibility above (below) its mean could potentially result in a 38% decrease (62% increase) in the conditional odds of being a nonideator.

In summary, these results intimate that negative interpretation inflexibility (above and beyond interpretation bias and depression) can predict denial of suicidal ideation and the severity of ideation among potential ideators. As expected, strong evidence emerged that depression also has this predictive utility. Finally, support was found for the notion that negative interpretation bias has utility in predicting the severity of suicidal ideation among potential ideators. Overall, these results are consistent with Hypothesis 1’s assertion that negative interpretation inflexibility and negative interpretation bias are uniquely associated with suicidal ideation. These results are also noteworthy because predictors of whether an individual will deny suicidal ideation can be useful in clinical contexts given that people may have reasons (e.g., social desirability) to deny suicidal ideation despite experiencing it (Cukrowicz et al., 2013). The present research suggests that clinicians might be merited in wondering whether individuals who are high on negative interpretation inflexibility and/or depression may be experiencing suicidal ideation even when they deny it.

Network analysis

To investigate potential pathways by which inflexible and biased interpretations may be related to suicidal ideation and the beliefs that putatively inspire it, we examined an undirected partial correlation network derived from Phase 2 baseline data. Network analysis was employed because it can simultaneously evaluate relations between all three metrics derived from the BADE task and beliefs highlighted in multiple theories of suicide, reducing the probability of Type I errors.

Network estimation

The goldbricker function (networktools; Jones, 2020) suggested that none of the nodes in the resulting network (Fig. 1) were redundant. Suicidal ideation was positively related to burdensomeness, depression, hopelessness, psychache, thwarted belongingness, defeat/entrapment, and a lack of positive future expectations and was negatively related to positive/negative interpretation bias. Despite the lack of a direct path between negative interpretation inflexibility and suicidal ideation, large positive edges connected negative interpretation inflexibility to suicidal ideation via burdensomeness. Given the nature of partial correlation networks, the presence of these edges is consistent with the assertion of Hypothesis 1 that beliefs identified in past theories of suicide will mediate any relation between interpretation inflexibility or bias and suicidal ideation.

Fig. 1.

Regularized partial correlation network. Annulus surrounding each node denotes predictability (more filled = more predictable). SSI = Scale for Suicide Ideation (Beck & Steer, 1991); Ache = psychache; BDI = Beck Depression Inventory (Beck et al., 1996); PRFT = pessimistic repetitive future thinking; PSS = Perceived Stress Scale; SDES = Short Defeat and Entrapment Scale (Griffiths et al., 2015); Burden = burdensomeness; BHS = Beck Hopeless Scale; NII = negative interpretation inflexibility; NIB = negative interpretation bias; PIB = positive interpretation bias; Belong = thwarted belongingness; NoPosFuture = confident anticipation of lack of positive future events; NegFuture = confident anticipation of negative future events.

Network inference

There were several significant differences among edge weights in the network (see Fig. S3, top, in the Supplemental Material). Note that edges leading from negative interpretation inflexibility to burdensomeness and from burdensomeness to suicidal ideation were statistically stronger than the majority of edges. Other edges that were statistically stronger than most others included those between hopelessness and defeat/entrapment, stress and pessimistic repetitive future thinking, and depression and psychache.

Significant differences were also observed between nodes’ expected influence (Fig. 2; also see Fig. S3, bottom, in the Supplemental Material). Defeat/entrapment had a statistically greater expected influence than every node. The expected influences of burdensomeness, psychache, and depression were stronger than the majority of other nodes. Negative interpretation inflexibility, negative interpretation bias, and positive interpretation bias had a smaller expected influence than many other nodes in the network.

Fig. 2.

Centrality measures (one-step expected influence) of nodes in regularized partial correlation network. SSI = Scale for Suicide Ideation (Beck & Steer, 1991); Ache = psychache. BDI = Beck Depression Inventory (Beck et al., 1996); PRFT = pessimistic repetitive future thinking; PSS = Perceived Stress Scale (Cohen, 1994); SDES = Short Defeat and Entrapment Scale (Griffiths et al., 2015); Burden = burdensomeness; BHS = Beck Hopelessness Scale (Beck et al., 1974); NII = negative interpretation inflexibility; NIB = negative interpretation bias; PIB = positive interpretation bias; Belong = thwarted belongingness; NoPosFuture = confident anticipation of lack of positive future events; NegFuture = confident anticipation of negative future events.

There was no correlation between the standard deviation of a node and its expected influence, ρ(12) = 0.33, p = .246. There was also no correlation between node means and their expected influences, ρ(12) = 0.11, p = .716. Thus, the relative average severity of beliefs related to suicide and differential node informativeness were not capable of explaining the centrality of nodes in the network.

Variance in nodes was generally well explained by their neighbors. The average predictability across nodes was .57, indicating that more than half of the variance in the network could be accounted for. However, 43% of the variance in the network remained unexplained and may be attributable to unmeasured variables. Defeat/entrapment (predictability = .82) was the best explained node in the network, followed by psychache (predictability = .74) and depression (predictability = .70). Approximately half of the variance was accounted for in suicidal ideation (predictability = .52) and negative interpretation inflexibility (predictability = .52).

Network accuracy and stability

Bootstrapped CIs suggested that the precision of network edges was acceptable (see Fig. S2 in the Supplemental Material). Edges were sufficiently stable, with a correlation-stability coefficient of .67, as was the one-step expected influence (CS-coefficient = .67). Consequently, findings regarding these metrics could be interpreted.

Potential mediators of longitudinal relations between interpretation bias/inflexibility and suicidal ideation

As a further test of Hypothesis 1 and to test Hypothesis 2, the possibility that the relation between negative interpretation inflexibility and suicidal ideation might be mediated by beliefs highlighted in major theories of suicide was examined in data from the longitudinal portion of the present study. This examination was carried out using three mediation models, each examining potential mediators from a different theory of suicide (while statistically controlling for depression). Only beliefs highlighted in theories employing ideation-to-action frameworks were examined because such theories explicitly identify specific predictors of suicidal ideation (and may better avoid confounding these predictors with those of suicidal behavior).

The first model examined beliefs (perceived burdensomeness and thwarted belongingness) from the interpersonal theory of suicide (Joiner, 2005) and depression as potential parallel mediators of the relation between negative interpretation inflexibility (and positive/negative interpretation bias) and suicidal ideation. This model was an excellent fit to the data: CFI = 1, SRMR = .00, RMSEA = 0.00. The direct path (path a1) from negative interpretation inflexibility (baseline) to perceived burdensomeness (Week 1) was significant, β = 0.15, 95% CI = [0.04, 0.25], z = 2.70, p > |z| = .007. The direct path (path b1) from perceived burdensomeness (Week 1) to suicidal ideation (Week 2) was also significant, β = 0.25, 95% CI = [0.03, 0.48], z = 2.18, p > |z| = .029. When these direct effects were considered jointly, the resulting indirect effect on suicidal ideation (Week 2) was statistically significant, β = 0.04, 95% CI = [0.02, 0.05]. In contrast, the indirect effects of negative interpretation inflexibility (baseline) on suicidal ideation (Week 2) via thwarted belongingness (Week 1; β = 0.00, 95% CI = [−0.01, 0.01]) and depression (Week 1; β = 0.00, 95% CI = [−0.01, 0.01]) were not significant. When controlling for these indirect effects, the remaining direct effect (path c’) of negative interpretation inflexibility (baseline) on suicidal ideation (Week 2) was not significant, β = 0.03, 95% CI = [−0.09, 0.16], z = 0.56, p > |z| = .576. The total effect combining this direct path and the aforementioned indirect pathway via perceived burdensomeness was significant, β = 0.06, 95% CI = [−0.05, 0.16]. These results are indicative of an indirect-only mediation pathway (see Zhao et al., 2010) leading from negative interpretation inflexibility to suicidal ideation via perceived burdensomeness. Table 2 contains all remaining parameter estimates.

Table 2.

Longitudinal Mediation Model: Interpersonal Theory of Suicide

Criterion and predictor	Estimate	SE	z	p
Regressions
SI (Week 2)
NII (baseline)	0.03 [−0.09, 0.16]	0.06	0.56	.576
NIB (baseline)	0.01 [−0.07, 0.09]	0.04	0.32	.750
PIB (baseline)	0.02 [−0.08, 0.11]	0.05	0.37	.712
PB (Week 1)	0.25 [0.03, 0.48]	0.12	2.18	.029
TB (Week 1)	−0.09 [−0.25, 0.06]	0.08	−1.21	.225
SI (baseline)	0.63 [0.47, 0.78]	0.08	7.83	< .001
PB (baseline)	−0.06 [−0.29 0.17]	0.12	−0.55	.583
TB (baseline)	0.04 [−0.11, 0.20]	0.08	0.53	.597
BDI (baseline)	−0.02 [−0.35, 0.31]	0.17	−0.11	.911
BDI (Week 1)	0.18 [−0.11, 0.47]	0.15	1.23	.220
PB (Week 1)
NII (baseline)	0.15 [0.04, 0.25]	0.05	2.70	.007
NIB (baseline)	0.02 [−0.04, 0.09]	0.03	0.69	.488
PIB (baseline)	0.02 [−0.08, 0.11]	0.05	0.30	.762
SI (baseline)	0.08 [−0.02, 0.17]	0.05	1.58	.114
PB (baseline)	0.66 [0.56, 0.77]	0.05	12.77	< .001
TB (baseline)	0.02 [−0.06, 0.10]	0.04	0.47	.640
BDI (baseline)	0.12 [0.00, 0.23]	0.06	2.00	.046
TB (Week 1)
NII (baseline)	−0.01 [−0.11, 0.09]	0.05	−0.14	.892
NIB (baseline)	0.00 [−0.11, 0.09]	0.04	0.00	.999
PIB (baseline)	−0.09 [−0.18, 0.00]	0.05	1.95	.051
SI (baseline)	−0.07 [0.16, 0.02]	0.05	−1.59	.112
PB (baseline)	0.10 [−0.01, 0.22]	0.06	1.73	.085
TB (baseline)	0.77 [0.69, 0.85]	0.04	17.98	< .001
BDI (baseline)	0.07 [−0.04, 0.19]	0.06	1.24	.215
BDI (Week 1)
NII (baseline)	−0.01 [−0.10 0.08]	0.05	−0.23	.821
NIB (baseline)	0.03 [−0.05, 0.10]	0.04	0.64	.522
PIB (baseline)	−0.04 [−0.13, 0.05]	0.04	−0.88	.380
SI (baseline)	−0.02 [−0.13, 0.09]	0.05	−0.38	.703
PB (baseline)	0.08 [−0.02, 0.18]	0.05	1.53	.126
TB (baseline)	0.01 [−0.06, 0.09]	0.04	0.40	.692
BDI (baseline)	0.84 [0.74, 0.93]	0.05	16.79	< .001
Intercepts
SI (Week 2)	0.00 [−0.08, 0.07]	0.36	−0.11	.915
PB (Week 1)	0.02 [−0.04, 0.08]	0.03	0.59	.556
TB (Week 1)	0.01 [−0.07, 0.08]	0.04	0.13	.894
BDI (Week 1)	0.02 [−0.04, 0.09]	0.03	0.72	.471

Note: The model is illustrated in Section S7 of the Supplemental Material available online. Values in brackets are 95% confidence intervals. Boldface type indicates significant results. SI = suicidal ideation; TB = thwarted belongingness; PB = perceived burdensomeness; NII = negative interpretation inflexibility; NIB = negative interpretation bias; PIB = positive interpretation bias; BDI = Beck Depression Inventory (Beck et al., 1996).

In accordance with previous studies examining suicidal ideation as an outcome in mediation analyses (Chu et al., 2018), zero-inflated negative binomial regressions were examined for consistency with the aforementioned model (see Section S11 in the Supplemental Material). These regressions provided additional support for the indirect pathway from negative interpretation inflexibility to suicidal ideation via perceived burdensomeness.

Models featuring beliefs highlighted in the integrated motivational and volitional model of suicidal behavior (R. C. O’Connor et al., 2016; R. C. O’Connor & Kirtley, 2018) and the three-step theory of suicide (Klonsky & May, 2015) were also examined. The direct effect of negative interpretation inflexibility on suicidal ideation was significant in the latter model and was marginally so in the former model. However, no indirect effects via beliefs (e.g., hopelessness, defeat/entrapment) relevant to suicidal ideation were significant. Results obtained using these models are fully described in Section S7 in the Supplemental Material.

Discussion

The present investigation builds on previous studies linking interpretation bias to suicidal ideation (Beard et al., 2017; Beevers & Miller, 2004) by using the emotional BADE task to simultaneously examine, for the first time, both biased and inflexible interpretations in relation to suicidal ideation and beliefs highlighted in major theories of suicide. Several key findings emerged from this investigation: In zero-inflated regression models of cross-sectional data, negative interpretation bias and (marginally) negative interpretation inflexibility were associated with suicidal ideation. Network analysis of cross-sectional data revealed a potential indirect pathway leading from negative interpretation inflexibility to suicidal ideation via perceived burdensomeness (which, notably, was present in several subgroups of participants, indicating the robust nature of this result: see Figs. S4 and S5 in the Supplemental Material). These findings hold even when statistically controlling for depression, which has previously been linked to inflexible negative interpretations (Everaert, Bronstein, et al., 2020; Everaert et al., 2018). These results strongly support the claims of Hypotheses 1 and 2 with respect to negative interpretation inflexibility but provide substantially less support for these claims with respect to interpretation bias.

These results synergize with extant research helping to clarify the relations between suicidal ideation, perceived burdensomeness, and thwarted belongingness (for a review, see Chu et al., 2017). Previous studies investigating these constructs indicate that perceived burdensomeness is the stronger predictor of suicidal ideation (Ma et al., 2016). In accordance with this previous research, the present study found that perceived burdensomeness, but not thwarted belongingness, was longitudinally related to suicidal ideation. Note that perceptions of burdensomeness among suicide ideators are often incorrect (Hames et al., 2013) and may be systematically distorted by information processing biases (as beliefs often are in the general population; see Nickerson, 1998). Consistent with this notion, in the present study, a stronger bias against revising negative interpretations of ambiguous social situations in response to positive disconfirmatory evidence longitudinally predicted greater endorsement of beliefs regarding one’s status as a burden to others and, in turn, suicidal ideation. This indirect pathway was also observed in the psychometric network we constructed from cross-sectional data.

The identification of this pathway provides the basis for future studies investigating whether inflexible negative interpretations cause beliefs about burdensomeness to persist, potentially encouraging interpersonal hopelessness (see Mandracchia et al., 2019) and suicidal ideation. Individuals with adverse experiences would be an interesting population for such studies. Adverse experiences, including discrimination (Sutter & Perrin, 2016), financial hardship (Economou et al., 2016), and military sexual assault (Monteith et al., 2019; Wiblin et al., 2021), have been linked to suicidal ideation. Although suicidal ideation in some individuals with these experiences may be fully driven by entirely accurate beliefs that are optimally updated, we posit that in many individuals, suicidal ideation after adverse experiences may be exacerbated by insufficient updating of beliefs that do not fully reflect reality—this theoretical position accords with research suggesting that human cognition is generally organized to resist belief updating (Altemeyer, 2002; Kaplan et al., 2016), particularly with respect to personally meaningful beliefs. For example, female veterans who experience military sexual assault may incorrectly blame themselves for their trauma, may falsely believe they are unlovable because of it, and may misperceive themselves as a burden to others, all of which may inspire suicidal ideation (Monteith et al., 2017, 2019; Wiblin et al., 2021). Insufficient updating of negative interpretations regarding ambiguous social interactions may lead these inaccurate beliefs to persist (e.g., not updating these interpretations to account for evidence that close others want to help an individual cope with trauma may result in the individual maintaining misperceptions of burdensomeness) and may thereby promote suicidal ideation. Future research should investigate this possibility.

The relations between inflexible negative interpretations, perceived burdensomeness, and suicidal ideation observed herein may also have clinical implications. Interventions targeting suicidal ideation have sometimes employed cognitive bias modification procedures aimed at altering biased interpretations that may encourage perceived burdensomeness (Allan et al., 2018). The present study suggests that these interventions might be augmented by incorporating modules from other therapies (e.g., the module on changing beliefs from metacognitive training for psychosis; Moritz et al., 2014) that likely encourage interpretation flexibility (see Buonocore et al., 2015).

The present study’s results are also relevant to several other theories of suicide. Much of this relevance stems from the fact that in the present study, in keeping with previous research examining suicidal ideation (de Beurs et al., 2019), we selected variables for network analysis on the basis of psychological theory. With this in mind, the fact that defeat/entrapment, which forms the backbone of the integrated motivational-volitional model of suicidal behavior (R. C. O’Connor & Kirtley, 2018), was the most central node in this network is significant. Although the centrality of defeat/entrapment cannot be interpreted as strong evidence that it plays an important causal role in this network (see Dablander & Hinne, 2019), it is by definition reflective of the extensive relations between defeat/entrapment and constructs from other theories of suicide (i.e., hopelessness, perceived burdensomeness, and psychache). The existence of these relations should encourage future research examining defeat/entrapment in the context of constructs from other theories of suicide; such research would be extremely beneficial given that studies examining multiple theories of suicide (like the present one) have been exceedingly rare (Kleiman et al., 2014) but are essential for determining which constructs are causally primary in pathways leading to suicide and suicidal ideation.

Beyond their relevance to various theories of suicidality, the present results are notable because they interface with the small but growing literature examining interpretation bias in the context of suicidal ideation. Two previous studies on this topic exist. The first found that negative interpretation bias longitudinally predicted suicidal ideation 6 months later via its effect on hopelessness (Beevers & Miller, 2004), and the second found that a bias against positive interpretations longitudinally predicts suicidal ideation over approximately 2 weeks (Beard et al., 2017). Both of these studies employed psychiatric hospital samples. In the general population sample employed in the present study, however, no evidence of a longitudinal relation between positive/negative interpretation bias and suicidal ideation over 2 weeks was observed. One potential explanation for the discrepancy between this and previous research is that the effects of interpretation bias on suicidal ideation may take longer than 2 weeks to become apparent in general population samples (which may feature less severe interpretation bias than samples of hospital patients). Future research investigating the relation between interpretation bias and suicidal ideation in the general population may therefore benefit from a longer follow-up period. This future research may wish to focus on the bias toward negative (rather than against positive) interpretations because this bias was more consistently related (cross-sectionally) to suicidal ideation and associated beliefs (e.g., hopelessness) in the present study.

A final noteworthy aspect of the present study is that it is the first to apply a scoring technique to the emotional BADE task that is not based on PCA. Evidence for the validity of this new strategy (which was first applied to the original BADE task in the context of research on psychosis; Sanford et al., 2014) was obtained through the replication of observations from previous studies employing the emotional BADE task (e.g., the relation between negative interpretation inflexibility, dampening of positive emotions, and depression; Everaert, Bronstein, et al., 2020), as will be detailed in a forthcoming manuscript. This scoring strategy confers several important advantages. For instance, it allows the emotional BADE task to be employed with small samples, which are incompatible with PCA but common in psychopathology research and clinical settings, and makes comparison of emotional BADE task scores across studies considerably easier. It is therefore expected that the application of this scoring strategy to the emotional BADE task will support continued research on interpretation inflexibility (and bias) in the context of suicidal thoughts and behaviors (and depression and social anxiety; see Everaert et al., 2018).

These implications of the present study should be considered in the context of several limitations. First, the present study’s network analyses were constructed from a cross-sectional data set and undirected graphs were produced, rendering us unable to discern the direction of any causal relation between variables in the networks. This limitation is substantially mitigated by the longitudinal models examined herein. However, in the interest of avoiding Type I error, not every relation present in the network was investigated longitudinally. A second limitation of the present study is that we examined constructs from important theories of suicide in relation to a measure of suicidal ideation (the SSI-5), which combines several aspects of suicidality (e.g., motivational factors, active and passive suicidal desire). Several of these constructs are theorized to be differentially related to various aspects of suicidality. For example, the interpersonal theory of suicide specifies that elevated levels of either perceived burdensomeness or thwarted belongingness should result in passive suicidal ideation, whereas the presence of both of these variables in addition to interpersonal hopelessness should result in active suicidal desire (Van Orden et al., 2010). By employing a multifaceted measure of suicidality, we may therefore have underestimated the association between suicidal ideation and constructs such as perceived burdensomeness or thwarted belongingness. In light of this limitation, future research might adopt more fine-grained measures of suicidality (e.g., measures focusing either on passive or active suicidal ideation). A third limitation of the present study was that a comparison of study dropouts and completers indicated that individuals who completed the study exhibited less interpretation inflexibility (see Section S4 in the Supplemental Material), raising the possibility that the results of longitudinal analyses in the present study involving this construct may not generalize as well to more inflexible individuals. Future research could address this limitation by repeating these analyses in a sample of individuals likely to exhibit more inflexible interpretations (e.g., depressed inpatients).

Although not limitations per se, two final points should be kept in mind when assessing the implications of this study. First, exploratory analyses detailed in Section S8 in the Supplemental Material indicated that inflexible positive interpretations, such as inflexible negative ones, were related to suicidal ideation (both cross-sectionally and over time) in a manner mediated by perceived burdensomeness. Accordingly, inflexible interpretations in general, rather than inflexible positive or negative interpretations specifically, may be the most relevant to suicidal ideation. Second, data for the present study were collected during the early phases of a global pandemic, and participants may have been subject to measures (e.g., physical and social distancing) meant to protect public health (for details on the timing of data collection and an extended discussion of the potential effects of this context, see Section S1 in the Supplemental Material). Because the effects of these measures and the pandemic itself on the relations between variables in the present study are unclear, it is possible that this study’s results might not generalize perfectly to populations free from such stressors, although this study’s replication of research conducted before the pandemic (see Section S1 in the Supplemental Material) tempers this concern somewhat. If future research reveals that pandemics and/or their associated public health measures result in unique relations between variables relevant to suicidal ideation, then the present study affords a unique window into these relations that may prove valuable, particularly because pandemics and their sequalae (e.g., quarantine) have been linked to increases in experiences associated with suicidality (e.g., stress and depressive symptoms; Liu et al., 2012).

Conclusion

Despite these considerations, the value of the present study is apparent against the backdrop of data—for example, the fact that suicide causes more than 800,000 deaths per year (World Health Organization, 2019) and is not strongly predicted by established risk factors in most existing research studies (Franklin et al., 2017)—implying that the current understanding of suicide is inadequate for the development of sufficiently effective suicide-prevention strategies. To the extent that identifying causes of suicidal ideation can inform interventions on suicide (see Jobes & Joiner, 2019), the present study may help ameliorate this tragic situation through its suggestion that inflexible interpretations may encourage beliefs regarding burdensomeness and, in turn, suicidal ideation. By identifying this potential mechanistic pathway, the present study may also scaffold improvements on existing treatments targeting suicidal ideation, which is strikingly common (Jobes & Joiner, 2019) and can cause distress and functional impairment (Van Spijker et al., 2014), even in the absence of suicidal behavior. Moreover, the present study is valuable because it is the first to simultaneously examine biased and inflexible interpretations in relation to suicidal ideation and because it provides exciting new directions for research examining how beliefs highlighted in prominent theories of suicide might be distorted by biases in information processing, such as the bias against revising interpretations in response to evidence against them.

Supplemental Material

sj-pdf-1-cpx-10.1177_2167702621993867 – Supplemental material for Inflexible Interpretations of Ambiguous Social Situations: A Novel Predictor of Suicidal Ideation and the Beliefs That Inspire It

Supplemental material, sj-pdf-1-cpx-10.1177_2167702621993867 for Inflexible Interpretations of Ambiguous Social Situations: A Novel Predictor of Suicidal Ideation and the Beliefs That Inspire It by Jonas Everaert, Michael V. Bronstein, Tyrone D. Cannon, E. David Klonsky and Jutta Joormann in Clinical Psychological Science

Footnotes

Transparency

Action Editor: John Curtin

Editor: Kenneth J. Sher

Author Contributions

J. Everaert and M. V. Bronstein contributed equally to this work. J. Everaert and M. V. Bronstein developed the study concept, which J. Joormann and E. D. Klonsky helped to refine. All authors contributed to the study design. M. V. Bronstein collected the data. J. Everaert and M. V. Bronstein drafted the manuscript. T. D. Cannon, J. Joormann, and E. D. Klonsky provided critical revisions. All of the authors approved the final manuscript for submission.

ORCID iD

Michael V. Bronstein

Supplemental Material

Additional supporting information can be found at

References

Abramson

L. Y.

Alloy

L. B.

Hogan

M. E.

Whitehouse

W. G.

Gibb

B. E.

Hankin

B. L.

Cornette

M. M.

(2006). The hopelessness theory of suicidality. In Joiner

Rudd

M. D.

(Eds.), Suicide science (pp. 17–32). Kluwer Academic Publishers. https://doi.org/10.1007/0-306-47233-3_3

Abramson

L. Y.

Metalsky

G. I.

Alloy

L. B.

(1989). Hopelessness depression: A theory-based subtype of depression. Psychological Review, 96(2), 358–372. https://doi.org/10.1037/0033-295X.96.2.358

Allan

N. P.

Boffa

J. W.

Raines

A. M.

Schmidt

N. B.

(2018). Intervention related reductions in perceived burdensomeness mediates incidence of suicidal thoughts. Journal of Affective Disorders, 234, 282–288. https://doi.org/10.1016/j.jad.2018.02.084

Altemeyer

(2002). Dogmatic behavior among students: Testing a new measure of dogmatism. Journal of Social Psychology, 142(6), 713–721. https://doi.org/10.1080/00224540209603931

Beard

Rifkin

L. S.

Björgvinsson

(2017). Characteristics of interpretation bias and relationship with suicidality in a psychiatric hospital sample. Journal of Affective Disorders, 207, 321–326. https://doi.org/10.1016/j.jad.2016.09.021

Beck

A. T.

Kovacs

Weissman

(1979). Assessment of suicidal intention: The Scale for Suicide Ideation. Journal of Consulting and Clinical Psychology, 47(2), 343–352. https://doi.org/10.1037/0022-006X.47.2.343

Beck

A. T.

Steer

R. A.

(1991). Manual for Beck Scale for Suicide Ideation. Psychological Corp.

Beck

A. T.

Steer

R. A.

Brown

G. K.

(1996). Manual for the Beck Depression Inventory-II. Psychological Corp.

Beck

A. T.

Weissman

Lester

Trexler

(1974). The measurement of pessimism: The Hopelessness Scale. Journal of Consulting and Clinical Psychology, 42(6), 861–865. https://doi.org/10.1037/h0037562

10.

Beevers

C. G.

Miller

I. W.

(2004). Perfectionism, cognitive bias, and hopelessness as prospective predictors of suicidal ideation. Suicide and Life-Threatening Behavior, 34(2), 126–137.

11.

Bentler

P. M.

Bonett

D. G.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88(3), 588–606. https://doi.org/10.1037/0033-2909.88.3.588

12.

Blanchette

Richards

(2010). The influence of affect on higher level cognition: A review of research on interpretation, judgement, decision making and reasoning. Cognition and Emotion, 24(4), 561–595. https://doi.org/10.1080/02699930903132496

13.

Bloch-Elkouby

Gorman

Schuck

Barzilay

Calati

Cohen

L. J.

Begum

Galynker

(2020). The suicide crisis syndrome: A network analysis. Journal of Counseling Psychology, 67(5), 595–607. https://doi.org/10.1037/cou0000423

14.

Bronstein

M. V.

Cannon

T. D.

(2018). Measuring bias against disconfirmatory evidence: An evaluation of BADE task scoring methods and the case for a novel method. Psychiatry Research, 261, 535–540. https://doi.org/10.1016/j.psychres.2018.01.007

15.

Bronstein

M. V.

Pennycook

Bear

Rand

D. G.

Cannon

T. D.

(2019). Belief in fake news is associated with delusionality, dogmatism, religious fundamentalism, and reduced analytic thinking. Journal of Applied Research in Memory and Cognition, 8(1), 108–117. https://doi.org/10.1016/j.jarmac.2018.09.005

16.

Buitron

Hill

R. M.

Pettit

J. W.

Green

K. L.

Hatkevich

Sharp

(2016). Interpersonal stress and suicidal ideation in adolescence: An indirect association through perceived burdensomeness toward others. Journal of Affective Disorders, 190, 143–149. https://doi.org/10.1016/j.jad.2015.09.077

17.

Buonocore

Bosia

Riccaboni

Bechi

Spangaro

Piantanida

Cocchi

Guglielmino

Bianchi

Smeraldi

Cavallaro

(2015). Combined neurocognitive and metacognitive rehabilitation in schizophrenia: Effects on bias against disconfirmatory evidence. European Psychiatry, 30(5), 615–621. https://doi.org/10.1016/j.eurpsy.2015.02.006

18.

Chandler

Shapiro

(2016). Conducting clinical research using crowdsourced convenience samples. Annual Review of Clinical Psychology, 12, 53–81. https://doi.org/10.1146/annurev-clinpsy-021815-093623

19.

Chen

(2008). Extended Bayesian information criteria for model selection with large model spaces. Biometrika, 95(3), 759–771. https://doi.org/10.1093/biomet/asn034

20.

Chu

Buchman-Schmitt

J. M.

Stanley

I. H.

Hom

M. A.

Tucker

R. P.

Hagan

C. R.

Rogers

M. L.

Podlogar

M. C.

Chiurliza

Ringer

F. B.

Michaels

M. S.

Patros

C. H. G.

Joiner

T. E.

Jr. (2017). The interpersonal theory of suicide: A systematic review and meta-analysis of a decade of cross-national research. Psychological Bulletin, 143(12), 1313–1345. https://doi.org/10.1037/bul0000123

21.

Chu

Walker

Stanley

Hirsch

Greenberg

Rudd

Joiner

(2018). Perceived problem-solving deficits and suicidal ideation: Evidence for the explanatory roles of thwarted belongingness and perceived burdensomeness in five samples carol. Journal of Personality and Social Psychology, 115(1), 137–160.

22.

Cohen

(1994). Perceived stress scale. Psychology, 1–3.

23.

Cohen

Kamarck

Mermelstein

(1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 385–396.

24.

Cole

A. B.

Wingate

L. R. R.

Tucker

R. P.

Rhoades-Kerswill

O’Keefe

V. M.

Hollingsworth

D. W.

(2015). The differential impact of brooding and reflection on the relationship between perceived stress and suicide ideation. Personality and Individual Differences, 83, 170–173. https://doi.org/10.1016/j.paid.2015.04.013

25.

Cox

L. A.

(2018). Modernizing the Bradford Hill criteria for assessing causal relationships in observational data. Critical Reviews in Toxicology, 48(8), 682–712. https://doi.org/10.1080/10408444.2018.1518404

26.

Cukrowicz

K. C.

Jahn

D. R.

Graham

R. D.

Poindexter

E. K.

Williams

R. B.

(2013). Suicide risk in older adults: Evaluating models of risk and predicting excess zeros in a primary care sample. Journal of Abnormal Psychology, 122(4), 1021–1030. https://doi.org/10.1037/a0034953

27.

Dablander

Hinne

(2019). Node centrality measures are a poor substitute for causal inference. Scientific Reports, 9(1), Article 6846. https://doi.org/10.1038/s41598-019-43033-9

28.

de Beurs

Fried

E. I.

Wetherall

Cleare

O’Connor

D. B.

Ferguson

O’Carroll

R. E.

O’Connor

R. C.

(2019). Exploring the psychology of suicidal ideation: A theory driven network analysis. Behaviour Research and Therapy, 120, Article 103419. https://doi.org/10.1016/j.brat.2019.103419

29.

de Beurs

D. P.

(2017). Network analysis: A novel approach to understand suicidal behaviour. International Journal of Environmental Research and Public Health, 14(3), Article 219. https://doi.org/10.3390/ijerph14030219

30.

de Beurs

D. P.

Fokkema

de Groot

M. H.

de Keijser

Kerkhof

(2015). Longitudinal measurement invariance of the Beck Scale for Suicide Ideation. Psychiatry Research, 225(3), 368–373. https://doi.org/10.1016/J.PSYCHRES.2014.11.075

31.

Desseilles

Perroud

Guillaume

Jaussent

Genty

Malafosse

Courtet

(2012). Is it valid to measure suicidal ideation by depression rating scales? Journal of Affective Disorders, 136, 398–404. https://doi.org/10.1016/j.jad.2011.11.013

32.

Dhingra

Klonsky

E. D.

Tapola

(2019). An empirical test of the three-step theory of suicide in U.K. university students. Suicide and Life-Threatening Behavior, 49(2), 478–487. https://doi.org/10.1111/sltb.12437

33.

Dozois

D. J. A.

Dobson

K. S.

Ahnberg

J. L.

(1998). A psychometric evaluation of the Beck Depression Inventory-II. Psychological Assessment, 10(2), 83–89. https://doi.org/10.1037/1040-3590.10.2.83

34.

Economou

Angelopoulos

Peppou

L. E.

Souliotis

Stefanis

(2016). Suicidal ideation and suicide attempts in Greece during the economic crisis: An update. World Psychiatry, 15(1), 83–84. https://doi.org/10.1002/wps.20296

35.

Elliott

Jones

P. J.

Schmidt

(2020). Central symptoms predict posttreatment outcomes and clinical impairment in anorexia nervosa: A network analysis. Clinical Psychological Science, 8(1), 139–154. https://doi.org/10.1177/2167702619865958

36.

Enders

C. K.

Bandalos

D. L.

(2001). The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling, 8(3), 430–457. https://doi.org/10.1207/S15328007SEM0803_5

37.

Epskamp

(2020). bootnet: Bootstrap methods for various network estimation routines (Version 1.3) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=bootnet

38.

Epskamp

Costantini

Haslbeck

Isvoranu

(2020). qgraph: Graph plotting methods, psychometric data visualization and graphical model estimation (Version 1.6.5) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=qgraph

39.

Epskamp

Borsboom

Fried

E. I.

(2018). Estimating psychological networks and their accuracy: A tutorial paper. Behavior Research Methods, 50(1), 195–212. https://doi.org/10.3758/s13428-017-0862-1

40.

Everaert

Bernstein

Joormann

Koster

E. H. W.

(2020). Mapping dynamic interactions among cognitive biases in depression. Emotion Review, 12(2), 93–110. https://doi.org/10.1177/1754073919892069

41.

Everaert

Bronstein

M. V.

Cannon

T. D.

Joormann

(2018). Looking through tinted glasses: Depression and social anxiety are related to both interpretation biases and inflexible negative interpretations. Clinical Psychological Science, 6(4), 517–528. https://doi.org/10.1177/2167702617747968

42.

Everaert

Bronstein

M. V.

Castro

Cannon

T. D.

Joormann

(2020). When negative interpretations persist, positive emotions don’t! Inflexible negative interpretations encourage depression and social anxiety by dampening positive emotions. Behaviour Research and Therapy, 124, Article 103510. https://doi.org/10.1016/j.brat.2019.103510

43.

Everaert

Podina

I. R.

Koster

E. H. W.

(2017). A comprehensive meta-analysis of interpretation biases in depression. Clinical Psychology Review, 58, 33–48. https://doi.org/10.1016/j.cpr.2017.09.005

44.

Forkmann

Teismann

Stenzel

J.-S.

Glaesmer

De Beurs

(2018). Defeat and entrapment: More than meets the eye? Applying network analysis to estimate dimensions of highly correlated constructs. BMC Medical Research Methodology, 18, Article 16. https://doi.org/10.1186/s12874-018-0470-5

45.

Foygel

Drton

(2010). Extended Bayesian information criteria for Gaussian graphical models. In Lafferty

Williams

Shawe-Taylor

Zemel

Culotta

Advances in neural information processing systems 23 (NIPS 2010) (pp. 604–612). Neural Information Processing Systems Foundation. https://papers.nips.cc/paper/2010/file/072b030ba126b2f4b2374f342be9ed44-Paper.pdf

46.

Foygel Barber

Drton

(2015). High-dimensional Ising model selection with Bayesian information criteria. Electronic Journal of Statistics, 9, 567–607. https://doi.org/10.1214/15-EJS1012

47.

Franklin

J. C.

Ribeiro

J. D.

Fox

K. R.

Bentley

K. H.

Kleiman

E. M.

Huang

Musacchio

K. M.

Jaroszewski

A. C.

Chang

B. P.

Nock

M. K.

(2017). Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin, 143(2), 187–232. https://doi.org/10.1037/bul0000084

48.

Fried

E. I.

Eidhof

Palic

Costantini

Huisman-van Dijk

Bockting

Engelhard

Armour

Nielsen

Karstoft

K. I.

(2018). Replicability and generalizability of posttraumatic stress disorder (PTSD) networks: A cross-cultural multisite study of PTSD symptoms in four trauma patient samples. Clinical Psychological Science, 6(3), 335–351. https://doi.org/10.1177/2167702617745092

49.

Fuller

W. A.

(1991). Simple estimators for the mean of skewed populations. Statistica Sinica, 1(1), 137–158.

50.

Greene

W. H.

(2012). Econometric analysis (71st ed.). Stern School of Business, New York University.

51.

Griffiths

A. W.

Wood

A. M.

Maltby

Taylor

P. J.

Panagioti

Tai

(2015). The development of the Short Defeat and Entrapment Scale (SDES). Psychological Assessment, 27(4), 1182–1194. https://doi.org/10.1037/pas0000110

52.

Hames

J. L.

Hagan

C. R.

Joiner

T. E.

(2013). Interpersonal processes in depression. Annual Review of Clinical Psychology, 9(1), 355–377. https://doi.org/10.1146/annurev-clinpsy-050212-185553

53.

Haslbeck

J. M. B.

(2019). mgm: Estimating time-varying k-order mixed graphical models (Version 1.2.7) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=mgm

54.

Haslbeck

J. M. B.

Fried

E. I.

(2017). How predictable are symptoms in psychopathological networks? A reanalysis of 18 published datasets. Psychological Medicine, 47(16), 2767–2776. https://doi.org/10.1017/S0033291717001258

55.

Heffer

Willoughby

(2017). A count of coping strategies: A longitudinal study investigating an alternative method to understanding coping and adjustment. PLOS ONE, 12(10), Article e0186057. https://doi.org/10.1371/journal.pone.0186057

56.

Henriksson

M. M.

Aro

H. M.

Marttunen

M. J.

Heikkinen

M. E.

Isometsa

E. T.

Kuoppasalmi

K. I.

Lonnqvist

J. K.

(1993). Mental disorders and comorbidity in suicide. American Journal of Psychiatry, 150(6), 935–940. https://doi.org/10.1176/ajp.150.6.935

57.

Holden

R. R.

Mehta

Cunningham

E. J.

McLeod

L. D.

(2001). Development and preliminary validation of a Scale of Psychache. Canadian Journal of Behavioural Science, 33(4), 224–232. https://doi.org/10.1037/h0087144

58.

L. T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

59.

Hubert

Van Der Veeken

(2008). Outlier detection for skewed data. Journal of Chemometrics, 22(3–4), 235–246. https://doi.org/10.1002/cem.1123

60.

Isometsä

(2014). Suicidal behaviour in mood disorders: Who, when, and why? Canadian Journal of Psychiatry, 59(3), 120–130. https://doi.org/10.1177/070674371405900303

61.

Jackman

(2020). pscl: Political science computational laboratory (Version 1.5.5) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=pscl

62.

Jobes

D. A.

Joiner

T. E.

(2019). Reflections on suicidal ideation. Crisis, 40(4), 227–230. https://doi.org/10.1027/0227-5910/a000615

63.

Joiner

T. E.

(2005). Why people die by suicide. Harvard University Press.

64.

Jones

(2020). networktools: Tools for identifying important nodes in networks (Version 1.2.2) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=networktools

65.

Jorgensen

T. D.

Pornprasertmanit

Schoemann

A. M.

Rosseel

(2019). semTools: Useful tools for structural equation modeling (Version 0.5.2) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=semTools

66.

Jyunn Lai

Chi Tan

Ting Wang

Chi Wu

Yi Wang

Chih Shen

(2018). Difference in cognitive flexibility between passive and active suicidal ideation in patients with depression. Neuropsychiatry, 8(4), 1182–1185. https://doi.org/10.4172/neuropsychiatry.1000446

67.

Kaliuzhna

Chambon

Franck

Testud

van der Henst

J. B.

(2012). Belief revision and delusions: How do patients with schizophrenia take advice? PLOS ONE, 7(4), Article e34771. https://doi.org/10.1371/journal.pone.0034771

68.

Kaplan

J. T.

Gimbel

S. I.

Harris

(2016). Neural correlates of maintaining one’s political beliefs in the face of counterevidence. Scientific Reports, 6, 1–11. https://doi.org/10.1038/srep39589

69.

Kirtley

O. J.

Melson

A. J.

O’Connor

R. C.

(2018). Future-oriented constructs and their role in suicidal ideation and enactment. In Hirsch

Chang.

Kelliher Rabon

(Eds.), A positive psychological approach to suicide. Advances in mental health and addiction (pp. 17–36). Springer. https://doi.org/10.1007/978-3-030-03225-8_2

70.

Kleiman

E. M.

Law

K. C.

Anestis

M. D.

(2014). Do theories of suicide play well together? Integrating components of the hopelessness and interpersonal psychological theories of suicide. Comprehensive Psychiatry, 55(3), 431–438. https://doi.org/10.1016/j.comppsych.2013.10.015

71.

Kliem

Lohmann

Mößle

Brähler

(2017). German Beck Scale for Suicide Ideation (BSS): Psychometric properties from a representative population survey. BMC Psychiatry, 17(1), Article 389. https://doi.org/10.1186/s12888-017-1559-9

72.

Klonsky

E. D.

May

A. M.

(2015). The Three-Step Theory (3ST): A new theory of suicide rooted in the “Ideation-to-Action” Framework. International Journal of Cognitive Therapy, 8(2), 114–129.

73.

Kocalevent

R. D.

Finck

Pérez-Trujillo

Sautier

Zill

Hinz

(2017). Standardization of the Beck Hopelessness Scale in the general population. Journal of Mental Health, 26(6), 516–522. https://doi.org/10.1080/09638237.2016.1244717

74.

Krajniak

Miranda

Wheeler

(2013). Rumination and pessimistic certainty as mediators of the relation between lifetime suicide attempt history and future suicidal ideation. Archives of Suicide Research, 17(3), 196–211. https://doi.org/10.1080/13811118.2013.805638

75.

Lervåg

(2019). Editorial: Correlation and causation: To study causality in psychopathology. Journal of Child Psychology and Psychiatry, 60(6), 603–605. https://doi.org/10.1111/jcpp.13074

76.

Liu

Kakade

Fuller

C. J.

Fang

Kong

Guan

(2012). Depression after exposure to stressful events: Lessons learned from the severe acute respiratory syndrome epidemic. Comprehensive Psychiatry, 53(1), 15–23. https://doi.org/10.1016/j.comppsych.2011.02.003

77.

Batterham

P. J.

Calear

A. L.

Han

(2016). A systematic review of the predictions of the Interpersonal-Psychological Theory of Suicidal Behavior. Clinical Psychology Review, 46, 34–45. https://doi.org/10.1016/j.cpr.2016.04.008

78.

MacKinnon

D. P.

Lockwood

C. M.

Hoffman

J. M.

West

S. G.

Sheets

(2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7(1), 83–104. https://doi.org/10.1037/1082-989X.7.1.83

79.

Mandracchia

J. T.

Sunderland

M. N.

Y. M.

(2019). Evaluating the role of interpersonal hopelessness in the interpersonal theory of suicide. Death Studies. Advance online publication. https://doi.org/10.1080/07481187.2019.1671549

80.

Marchetti

(2020). Hopelessness: A network analysis. Cognitive Therapy and Research, 43(3), 611–619.

81.

McDonald

(1999). Test theory: A unified treatment. Erlbaum.

82.

Meier

Kossakowski

J. J.

Jones

P. J.

Kay

Riemann

B. C.

McNally

R. J.

(2019). Obsessive–compulsive symptoms in eating disorders: A network investigation. International Journal of Eating Disorders, 53(3), 362–371. https://doi.org/10.1002/eat.23196

83.

Miller

A. B.

Prinstein

M. J.

(2019). Adolescent suicide as a failure of acute stress-response systems. Annual Review of Clinical Psychology, 15(1), 425–450. https://doi.org/10.1146/annurev-clinpsy-050718-095625

84.

Miranda

Gallagher

Bauchner

Vaysman

Marroquín

(2012). Cognitive inflexibility as a prospective predictor of suicidal ideation among young adults with a suicide attempt history. Depression and Anxiety, 29(3), 180–186. https://doi.org/10.1002/da.20915

85.

Miranda

Mennin

D. S.

(2007). Depression, generalized anxiety disorder, and certainty in pessimistic predictions about the future. Cognitive Therapy and Research, 31(1), 71–82. https://doi.org/10.1007/s10608-006-9063-4

86.

Miranda

Valderrama

Tsypes

Gadol

Gallagher

(2013). Cognitive inflexibility and suicidal ideation: Mediating role of brooding and hopelessness. Psychiatry Research, 210(1), 174–181. https://doi.org/10.1016/j.psychres.2013.02.033

87.

Miranda

Wheeler

Polanco-Roman

Marroquín

(2017). The Future-Oriented Repetitive Thought (FoRT) Scale: A measure of repetitive thinking about the future. Journal of Affective Disorders, 207, 336–345. https://doi.org/10.1016/j.jad.2016.09.055

88.

Monteith

L. L.

Bahraini

N. H.

Menefee

D. S.

(2017). Perceived burdensomeness, thwarted belongingness, and fearlessness about death: Associations with suicidal ideation among female veterans exposed to military sexual trauma. Journal of Clinical Psychology, 73(12), 1655–1669. https://doi.org/10.1002/jclp.22462

89.

Monteith

L. L.

Holliday

Schneider

A. L.

Forster

J. E.

Bahraini

N. H.

(2019). Identifying factors associated with suicidal ideation and suicide attempts following military sexual trauma. Journal of Affective Disorders, 252, 300–309. https://doi.org/10.1016/j.jad.2019.04.038

90.

Moritz

Andreou

Schneider

B. C.

Wittekind

C. E.

Menon

Balzan

R. P.

Woodward

T. S.

(2014). Sowing the seeds of doubt: A narrative review on metacognitive training in schizophrenia. Clinical Psychology Review, 34(4), 358–366. https://doi.org/10.1016/j.cpr.2014.04.004

91.

Nickerson

R. S.

(1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175–220. https://doi.org/10.1214/aos/1031594735

92.

O’Connor

D. B.

Green

J. A.

Ferguson

O’Carroll

R. E.

O’Connor

R. C.

(2017). Cortisol reactivity and suicidal behavior: Investigating the role of hypothalamic-pituitary-adrenal axis responses to stress in suicide attempters and ideators. Psychoneuroendocrinology, 75, 183–191. https://doi.org/10.1016/j.psyneuen.2016.10.019

93.

O’Connor

R. C.

(2011). The integrated motivational-volitional model of suicidal behavior. Crisis, 32(6), 295–298. https://doi.org/10.1027/0227-5910/a000120

94.

O’Connor

R. C.

Cleare

Eschle

Wetherall

Kirtley

O. J.

(2016). The integrated motivational-volitional model of suicidal behavior: An update. In O’Connor

R. C.

Pirkis

(Eds.), The international handbook of suicide prevention (2nd ed., pp. 220–240). Wiley Blackwell. https://doi.org/10.1002/9781118903223.ch13

95.

O’Connor

R. C.

Kirtley

O. J.

(2018). The integrated motivational-volitional model of suicidal behaviour. Philosophical Transactions of the Royal Society B: Biological Sciences, 373, Article 20170268. https://doi.org/10.1098/rstb.2017.0268

96.

O’Connor

R. C.

Portzky

(2018). The relationship between entrapment and suicidal behavior through the lens of the integrated motivational–volitional model of suicidal behavior. Current Opinion in Psychology, 22, 12–17. https://doi.org/10.1016/j.copsyc.2017.07.021

97.

Ophir

Sisso

Asterhan

C. S. C.

Tikochinski

Reichart

(2018). The Turker Blues: Hidden factors behind increased depression rates among Amazon’s Mechanical Turkers. Clinical Psychological Science, 8(1), 65–83. https://doi.org/10.1177/2167702619865973

98.

Prims

J. P.

Sisso

Bai

(2018). Flag suspects app [Computer software]. https://itaysisso.shinyapps.io/Bots

99.

R Core Team. (2019). R: A language and environment for statistical computing (Version 3.6.1) [Computer software]. R Foundation for Statistical Computing. http://www.R-project.org

100.

Ripley

(2020). MASS: Support functions and datasets for Venables and Ripley’s MASS (Version 7.3-51.6) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=MASS

101.

Roberti

J. W.

Harrington

L. N.

Storch

E. A.

(2006). Further psychometric support for the 10-item version of the Perceived Stress Scale. Journal of College Counseling, 9(2), 135–147. https://doi.org/10.1002/j.2161-1882.2006.tb00100.x

102.

Robinaugh

D. J.

Millner

A. J.

McNally

R. J.

(2016). Identifying highly influential nodes in the complicated grief network. Journal of Abnormal Psychology, 125(6), 747–757. https://doi.org/10.1037/abn0000181

103.

Rogers

M. L.

Joiner

T. E.

(2017). Rumination, suicidal ideation, and suicide attempts: A meta-analytic review. Review of General Psychology, 21(2), 132–142. https://doi.org/10.1037/gpr0000101

104.

Rosseel

(2019). lavaan: Latent variable analysis (Version 0.6-5) [Computer software]. Comprehensive R Archive Network. https://cran.r-project.org/package=lavaan

105.

Sanford

Veckenstedt

Moritz

Balzan

R. P.

Woodward

T. S.

(2014). Impaired integration of disambiguating evidence in delusional schizophrenia patients. Psychological Medicine, 44(13), 2729–2738. https://doi.org/10.1017/S0033291714000397

106.

Sargalska

Miranda

Marroquín

(2011). Being certain about an absence of the positive: Specificity in relation to hopelessness and suicidal ideation. International Journal of Cognitive Therapy, 4(1), 104–116. https://doi.org/10.1521/ijct.2011.4.1.104

107.

Shahnaz

Saffer

B. Y.

Klonsky

E. D.

(2018). The relationship of perfectionism to suicide ideation and attempts in a large online sample. Personality and Individual Differences, 130, 117–121. https://doi.org/10.1016/J.PAID.2018.04.002

108.

Shneidman

E. S.

(1993). Suicide as psychache: A clinical approach to self-destructive behavior. Aronson.

109.

Sturman

E. D.

(2011). Involuntary subordination and its relation to personality, mood, and submissive behavior. Psychological Assessment, 23(1), 262–276. https://doi.org/10.1037/a0021499

110.

Sutter

Perrin

P. B.

(2016). Discrimination, mental health, and suicidal ideation among LGBTQ people of color. Journal of Counseling Psychology, 63(1), 98–105. https://doi.org/10.1037/cou0000126

111.

Taylor

P. J.

Wood

A. M.

Gooding

Johnson

Tarrier

(2009). Are defeat and entrapment best defined as a single construct? Personality and Individual Differences, 47(7), 795–797. https://doi.org/10.1016/j.paid.2009.06.011

112.

Terluin

De Boer

M. R.

De Vet

H. C. W.

(2016). Differences in connection strength between mental symptoms might be explained by differences in variance: Reanalysis of network data did not confirm staging. PLOS ONE, 11(11), Article e0155205. https://doi.org/10.1371/journal.pone.0155205

113.

Tibshirani

(2016). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society. Series B (Methodological), 58(1), 267–288.

114.

Van Orden

K. A.

Cukrowicz

K. C.

Witte

T. K.

Joiner

T. E.

(2012). Thwarted belongingness and perceived burdensomeness: Construct validity and psychometric properties of the Interpersonal Needs Questionnaire. Psychological Assessment, 24(1), 197–215. https://doi.org/10.1037/a0025358

115.

Van Orden

K. A.

Witte

T. K.

Cukrowicz

K. C.

Braithwaite

S. R.

Selby

E. A.

Joiner

T. E.

(2010). The interpersonal theory of suicide. Psychological Review, 117(2), 575–600. https://doi.org/10.1037/a0018697

116.

Van Spijker

B. A. J.

Batterham

P. J.

Calear

A. L.

Farrer

Christensen

Reynolds

Kerkhof

A. J. F. M.

(2014). The Suicidal Ideation Attributes Scale (SIDAS): Community-based validation study of a new scale for the measurement of suicidal ideation. Suicide and Life-Threatening Behavior, 44(4), 408–419. https://doi.org/10.1111/sltb.12084

117.

Verrocchio

M. C.

Carrozzino

Marchetti

Andreasson

Fulcheri

Bech

(2016). Mental pain and suicide: A systematic review of the literature. Frontiers in Psychiatry, 7, Article 108. https://doi.org/10.3389/fpsyt.2016.00108

118.

Vuong

Q. H.

(1989). Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica, 57(2), 307–333.

119.

Vuorilehto

Valtonen

H. M.

Melartin

Sokero

Suominen

Isometsä

E. T.

(2014). Method of assessment determines prevalence of suicidal ideation among patients with depression. European Psychiatry, 29(6), 338–344. https://doi.org/10.1016/j.eurpsy.2013.08.005

120.

Wang

Y.-Y.

Jiang

N.-Z.

Cheung

E. F. C.

Sun

H.-W.

Chan

R. C. K.

(2015). Role of depression severity and impulsivity in the relationship between hopelessness and suicidal ideation in patients with major depressive disorder. Journal of Affective Disorders, 183, 83–89. https://doi.org/10.1016/j.jad.2015.05.001

121.

Ward

Peters

Jackson

Day

Garety

P. A.

(2018). Data-gathering, belief flexibility, and reasoning across the psychosis continuum. Schizophrenia Bulletin, 44(1), 126–136. https://doi.org/10.1093/schbul/sbx029

122.

Wiblin

Holder

Holliday

Surís

(2021). Predictors of unbearability, unlovability, and unsolvability in veterans with military-sexual-trauma-related posttraumatic stress disorder. Journal of Interpersonal Violence, 36(7–8), 3814–3830. https://doi.org/10.1177/0886260518777554

123.

Woodward

T. S.

Moritz

Menon

Klinge

(2008). Belief inflexibility in schizophrenia. Cognitive Neuropsychiatry, 13(3), 267–277. https://doi.org/10.1080/13546800802099033

124.

World Health Organization. (2019). Suicide in the world: Global health estimates. World Health Organization. https://apps.who.int/iris/handle/10665/326948

125.

Zhao

Lynch

J. G.

Chen

(2010). Reconsidering Baron and Kenny: Myths and truths about mediation analysis. Journal of Consumer Research, 37(2), 197–206. https://doi.org/10.1086/651257

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.38 MB